JP2000333175A - Image coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an image coder that can effectively compress a data quantity and reliably suppress visual deterioration of a reproduced image. SOLUTION: The image coder is provided with a data processing means consisting of a data processing circuit 9 that detects whether the pixel block of image data includes many high frequency components and performs control as to whether the AC coefficient of a preset part in a quantized discrete cosine transform coefficient Ruv is to be forcibly replaced with zero in response to a replacement control signal. When the data processing means discriminates that the high frequency components of the pixel block of the image data are comparatively small, the data of the part are abandoned by replacing some non-zero AC coefficients in the quantized discrete cosine transform coefficient Ruv by zero by the replacement control signal so as to reduce the compression data quantity, and when the data processing means discriminates that the high frequency components of the pixel block of the image data are comparatively large, the AC coefficient in the quantized discrete cosine transform coefficient Ruv is outputted as it is.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to image code amount control for effectively processing an effective coefficient value of a quantized discrete cosine transform (DCT) coefficient in an image encoding device. In particular, JPEG (Joint Photo)
The present invention relates to an image code amount control capable of performing effective data compression by omitting a high-frequency component that does not easily cause visual deterioration of a reproduced image in a data compression device using a graphic expert group (Graphic Expert Group) method.

[0002]

2. Description of the Related Art Conventionally, in a data compression apparatus using the JPEG system, Huffman coding is performed to reduce the code amount. However, since Huffman coding is a variable-length coding method, it has the property that the amount of coding differs depending on the type of image. That is, a large compression ratio can be obtained for a wall of one color or the sky, but a large compression ratio cannot be obtained for an image having a large change. Therefore, the number of pieces of image data that can be accommodated in one disc varies depending on the image.
In some cases, a single image can be accommodated, and other disks can accommodate only five images.

[0003] The present invention suppresses a change in the amount of coding, and
It is an object of the present invention to provide an image coding apparatus capable of storing a certain number of images on a single disk.

Before describing the present invention, a conventional encoding technique will be described. FIG. 5 is a block diagram showing a conventional image encoding device. In FIG. 5, 1 is an input terminal for inputting image data of an 8 × 8 pixel block, 2 is a discrete cosine transform unit that performs a discrete cosine transform (DCT), 3 is a zigzag transform unit, 4 is a quantizing unit, 5 is a quantum unit. Table,
Reference numeral 6 denotes an entropy coding unit, 7 denotes a coding table, and 8 denotes an output terminal for outputting parameters and code data.

Next, the operation will be described. First, as image data, a component image Pxy (x, y = 0, 1, 1, for example, having an 8-bit width) is input from the image data input terminal 1.
2, 3,..., 7) are input. The input image data is sent to the discrete cosine transform unit 2. In the discrete cosine transform unit 2, the divided 8 × 8 pixel block Px
Perform a two-dimensional discrete cosine transform on y.

As a result of the two-dimensional discrete cosine transform, 64 (= 8 × 8) coefficients Suv are obtained. The obtained 64 coefficients are rearranged from the serial order to the zigzag order by the zigzag converter 3 and sent to the quantizer 4. The 64 coefficients are quantized by the quantization unit 4 using the quantization table 5 with a different step size for each coefficient position. The 64 quantized coefficients are sent to the entropy coding unit 6.
The entropy coding unit 6 uses the coding table 7 to perform coding according to the Huffman coding method, and codes data from the output terminal 8 in units of several bytes (for example, 16-bit width).
Is output.

[0007] First, a component image Pxy (x, y = 0, 1, 2, 3, ..., 7) of 8 × 8 pixels is subjected to a two-dimensional discrete cosine transform in a discrete cosine transform unit 2, and the following ( The coefficient Suv as shown in the equation 1) is obtained.

[0008]

(Equation 1)

Here, the component image Pxy is represented by the following equation (2).

[0010]

(Equation 2)

The DCT coefficient Suv is, as shown in FIG.
S00 [DC (direct current) component] and other S01 to S7
And AC component (AC) up to 7 and a DC component S
00 is the maximum, and the other AC components are very small values compared to the S00 coefficient.

Next, the DCT coefficient Suv is converted from a serial order to a zigzag order by a zigzag converter 3 and then input to a quantizer 4. The DCT coefficient Suv is divided by the quantization unit 4 using the value Quv of the quantization table 5. That is, the quantized quantized DCT coefficient Ru
v is obtained by the following equation.

Ruv = round (Suv / Quv) Here, the round function is a function for converting to an integer closest to the calculation result of Suv / Quv. Therefore, by determining the value Quv of the quantization table so that the two-dimensional order uv becomes large where it is large, A
Most of the C coefficient can be set to 0 in a portion having a large order.

Next, the 64 coefficients Ruv output from the quantization unit 4 are sent to the entropy coding unit 6. In the entropy encoding unit 6, a DC coefficient (R00) representing an average value of 8 × 8 pixels and other AC coefficients (other than R00)
Since the encoding methods are different, they will be described separately.

First, a block diagram for grouping DC coefficients R00 is shown in FIG. In FIG. 7, the DC coefficient R00 input from the quantization unit 4 is
The difference between the DC coefficient R00 delayed by 1 and the immediately preceding DC coefficient R00 is subtracted by the DC differentiator 62, the difference is obtained, and the difference is sent to the grouping unit 63. As shown in FIG. 8, the arithmetic processing of the DC differentiator 62 is performed on the DC coefficient (DCi) of the current DC component block (i) and the block (i-1) of the same color component encoded immediately before. DC coefficient (DCi-
1) is calculated, and the difference value (ΔDCi) is obtained. Except for special images such as computer graphics, the average value with the adjacent block does not change much, so that the difference value from the immediately preceding DC coefficient is concentrated near zero. Therefore, by encoding the difference obtained in this way, highly efficient encoding can be expected.

The difference value of the DC coefficient obtained by the above-mentioned DC differentiator 62 is input to the grouping section 63,
Is used, the group to which the difference value belongs is obtained. The output from the grouping unit 63 is DC
The difference value is represented by a group number (S) and an additional bit (A). Here, the additional bit (A) is a value indicating the order of the difference value within the group. In FIG. 9, for example,
In group 3, the number of additional bits is 3, and DC
The difference value is -7, -6, -5, -4, 4, 5, 6, 7
Since it can take a value, 000 for -7, 001 for -6, -5
010 for -4, 011 for -4, 100 for 4 and 101 for 6
Are assigned to 110 and 7 to 111, respectively. As described above, the grouping unit 63 outputs the group number (S) and the additional bit (A). The group number (S) and the additional bit (A) output from the grouping unit 63 are one-dimensional Huffman encoded by the one-dimensional Huffman encoding unit 65 as described below.

FIG. 10 is a block diagram showing the grouping of AC coefficients. As shown in FIG. 11, the AC coefficients are output in a zigzag order because they have been rearranged by the zigzag converter 3. When the AC coefficients rearranged in the zigzag order are determined to be 0 by the determining unit 92, the run length counter 93 counts the number of consecutive AC coefficients 0 and outputs the run length (N).

When the AC coefficient is other than 0, the grouping unit 94
In, the group number (S) and the additional bit (A) are generated by the same method as that used to determine the DC difference value.
The group to which the AC coefficient belongs is determined using the table of FIG. Here, the additional bit is a value indicating the order of the AC coefficients in the group. For example, if the AC coefficient is 7, the group number is 3. In group 3, the number of additional bits is 3, and the DC difference value is -7,-
6, -8, -5, -4,4,5,6,7 are available, so 000 for -7, 001 for -6, 010 for -5, -4
011, 4 to 100, 5 to 101, 6 to 110, and 7 to 111. As described above, the grouping unit 94 outputs the group number (S) and the additional bit (A).

The group number (S) output from the grouping section 94 and the run length (N) output from the run length counter 93 are, as described later, the two-dimensional Huffman encoding section of the Huffman encoding section 70. At 95, Huffman coding is performed by the AC code table unit 96.

FIG. 13 is a block diagram showing a circuit for performing grouping and Huffman coding of DC coefficients and AC coefficients. In FIG. 13, the DC grouping unit 6
0a is the same as the circuit shown in FIG. 7, and the AC grouping unit 60b is the same as the circuit shown in FIG.

Hereinafter, the Huffman encoding unit 70 will be briefly described. In FIG. 13, a one-dimensional Huffman coding unit 65, a DC coding table unit 66, and a DC additional bit combining unit 67 are coding units for DC coefficients, and are one-dimensional Huffman coding unit 95, an AC coding table unit 96, The AC additional bit combining unit 97 is an encoding unit for AC coefficients. 68 is D
This is a combining circuit that combines the C-coded signal and the AC-coded signal.

First, coding of DC coefficients will be described.
Of the group number S and the additional bit A obtained by the DC grouping unit 60a, the group number S is one-dimensional Huffman encoded by the one-dimensional Huffman encoding unit 65 using the DC encoding table in the DC code table unit 66. And output as a DC code.

An example of one-dimensional Huffman coding is shown in FIG. For example, when the above group number 3 is input to the one-dimensional Huffman encoding unit 65, the one-dimensional Huffman encoded DC encoding becomes 110 from FIG. The output 110 output from the one-dimensional Huffman coding unit 65 is the additional bit 100 (4) output from the grouping unit 60a.
Is combined with the DC additional bit combining unit 67 to
(DC code + additional bit).

Next, encoding of AC coefficients will be described. In FIG. 13, the group number S output from the grouping unit 60b and the run length N output from the run length counter 93 of the grouping unit 60b are the same as the two-dimensional Huffman coding unit 95 of the Huffman coding unit 70 and the AC code. The Huffman coding is performed by the conversion table unit 96. The additional bit A is added to the encoded value by an AC additional bit combining unit 97, and the resultant value is output as an AC code. Two-dimensional Huffman coding will be described in a specific example below.

As shown in FIG. 15, following the DC coefficient,
AC coefficient (AC1, value = 3), 4 invalid coefficients of 0 value,
One AC coefficient (AC2, value = 10) and the remaining 57 are 0
Consider a case where a block signal of a value invalid coefficient is input to the circuit of FIG. 13 as an example.

According to the above example, first, the invalid coefficient 0 of 0 value
Are two-dimensional Huffman encoding units 9 as run lengths (N = 0).
5 is input. On the other hand, the next AC1 (value = 3) is 2
It is input to the dimensional Huffman encoding unit 95. Since the value of the AC coefficient is 3, the group number S is 2 from FIG. 12, and the value 3 of the effective coefficient is the maximum value of the group, so the additional bit is 11. Therefore, in the two-dimensional Huffman encoding unit 95, the invalid coefficient (N) is 0 and the group number (S) is 2 as described above, so that N / S is 0/2. Then, in the two-dimensional Huffman encoding section 95 of the Huffman encoding section 70, N / S
AC code table section 96 corresponding to (0/2) (FIG. 1)
6), two-dimensional Huffman coding is performed, additional bits are added thereto, and the two-dimensional Huffman coded signal "10
011 ”is obtained.

Next, the run length is 4 in the portion where there are four zero-value invalid coefficients and one AC coefficient (value = 10), while
From FIG. 14, the group number (S) is 4, and the additional bit is 1010 (10). Therefore, the N / S signal becomes 4/4, and two-dimensional Huffman encoding is performed using the AC encoding table 96 (FIG. 16), and an additional bit is added to the two-dimensional Huffman encoding signal to make the two-dimensional Huffman encoded signal “111111”.
11100110001010 "is obtained. For the next remaining 57 invalid coefficients of 0 value, EOB (End o)
f Block) code “00” is added, and the encoding of the block is completed.

Therefore, as described above, the Huffman coding method is a variable-length coding method in which the code length changes depending on data that appears, and the code amount changes depending on the original image.

[0029]

Since the conventional image encoding apparatus is configured as described above, the compression rate differs depending on the image, and the encoding amount (compressed data amount) is reduced even for images of the same size. There was a problem that it was very different.

The present invention has been made in order to solve the above-described problems, and it is possible to effectively compress the amount of data and to surely suppress visual deterioration of a reproduced image. It is intended to provide a device.

[0031]

In the image encoding apparatus according to the first aspect of the present invention, each component image Pxy (where the degree xy is a two-dimensional component of the image data of each component image) constituting an image block of the image data is provided. A discrete cosine transform is performed for each of the component images Pxy in the image data of each component image Pxy, where a discrete cosine transform is performed for each component image Pxy. A discrete cosine transform unit for deriving a two-dimensional position on the image data corresponding to an order xy indicating a two-dimensional position), and a discrete cosine transform unit comprising a coefficient having a DC component or an AC component derived by the discrete cosine transform unit. Cosine transform coefficient S
a quantizer for linearly quantizing uv at different step sizes for each two-dimensional position on the image data to derive a quantized discrete cosine transform coefficient Ruv, and the quantized discrete cosine arranged in a predetermined order and arranged in a serial order Conversion coefficient Ru
a zigzag transformation unit that rearranges v in a zigzag manner, and an entropy encoding unit that performs entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding method, and an image encoding device that compresses image data. Detecting whether or not a high-frequency component of a pixel block in the image data is large and controlling whether or not to output a replacement control signal, the AC coefficient of a predetermined portion of the quantized discrete cosine transform coefficient Ruv Data processing means for controlling whether to forcibly replace the pixel data with 0 in response to a replacement control signal, wherein the replacement is performed when the data processing means determines that the high frequency component of the pixel block in the image data is relatively small. Some AC coefficients in the quantized discrete cosine transform coefficient Ruv by the control signal for By replacing the non-zero part with zero, the data in that part is truncated to reduce the amount of compressed data, and when the data processing unit determines that the high frequency component of the pixel block in the image data is relatively large, the quantum The AC coefficients in the generalized discrete cosine transform coefficients Ruv are output as they are.

In the image encoding apparatus according to the second invention, each component image Pxy (where the degree xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data is provided. A discrete cosine transform is performed, and a discrete cosine transform coefficient Suv composed of coefficients having a DC component or an AC component arranged in a predetermined order (where the order uv is equal to each component image P
x representing the two-dimensional position of the xy in the image data
a discrete cosine transform unit for deriving a two-dimensional position on the image data corresponding to y), and a discrete cosine transform coefficient Suv composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit. A quantizer for performing linear quantization at a different step size for each two-dimensional position on image data to derive a quantized discrete cosine transform coefficient Ruv; and a quantized discrete cosine transform coefficient Ruv arranged in a predetermined order and arranged in a serial order. A zigzag transform unit that rearranges zigzag, and an entropy encoding unit that performs entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding method, wherein the image encoding device compresses image data. Detects and replaces whether high frequency components of pixel blocks in image data are large Controls or not issued or issues a control signal, said array of quantized discrete cosine transform coefficients Ruv
Data processing means for controlling whether or not AC coefficients after a predetermined specific coefficient are forcibly replaced with 0 in accordance with the replacement control signal, wherein the data processing means When it is determined that the high-frequency component is relatively small, the non-zero part of some AC coefficients in the quantized discrete cosine transform coefficient Ruv is replaced with zero by the replacement control signal, thereby truncating and compressing data of the part. The data amount is reduced, and the AC coefficient in the quantized discrete cosine transform coefficient Ruv is output as it is when the data processing means determines that the high frequency component of the pixel block in the image data is relatively large. is there.

In the image coding apparatus according to the third invention, each component image Pxy (where the degree xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data is provided. A discrete cosine transform is performed, and a discrete cosine transform coefficient Suv composed of coefficients having a DC component or an AC component arranged in a predetermined order (where the order uv is equal to each component image P
x representing the two-dimensional position of the xy in the image data
a discrete cosine transform unit for deriving a two-dimensional position on the image data corresponding to y), and a discrete cosine transform coefficient Suv composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit. A quantizer for performing linear quantization at a different step size for each two-dimensional position on image data to derive a quantized discrete cosine transform coefficient Ruv; and a quantized discrete cosine transform coefficient Ruv arranged in a predetermined order and arranged in a serial order. A zigzag transform unit that rearranges zigzag, and an entropy encoding unit that performs entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding method, wherein the image encoding device compresses image data. Detects and replaces whether high frequency components of pixel blocks in image data are large Controlling whether or not to output a control signal, and setting the AC coefficient of a predetermined portion of the quantized discrete cosine transform coefficients Ruv arranged in zigzag by the zigzag converter in accordance with the replacement control signal. A data processing means for controlling whether or not the high-frequency component of the pixel block in the image data is relatively small by the data processing means. A in the generalized discrete cosine transform coefficient Ruv
When the non-zero portion of the C coefficient is replaced with 0, the data in that portion is truncated to reduce the amount of compressed data. The AC coefficient in the quantized discrete cosine transform coefficient Ruv is directly output.

In the image coding apparatus according to the fourth invention, each component image Pxy (where the degree xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data is provided. A discrete cosine transform is performed, and a discrete cosine transform coefficient Suv composed of coefficients having a DC component or an AC component arranged in a predetermined order (where the order uv is equal to each component image P
x representing the two-dimensional position of the xy in the image data
a discrete cosine transform unit for deriving a two-dimensional position on the image data corresponding to y), and a discrete cosine transform coefficient Suv composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit. A quantizer for performing linear quantization at a different step size for each two-dimensional position on image data to derive a quantized discrete cosine transform coefficient Ruv; and a quantized discrete cosine transform coefficient Ruv arranged in a predetermined order and arranged in a serial order. A zigzag transform unit that rearranges zigzag, and an entropy encoding unit that performs entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding method, wherein the image encoding device compresses image data. Detects and replaces whether high frequency components of pixel blocks in image data are large Controlling whether or not to output a control signal, and replacing the AC coefficient after a predetermined coefficient among the quantized discrete cosine transform coefficients Ruv rearranged and arranged in a zigzag manner by the zigzag conversion unit with the replacement control signal A data processing unit for controlling whether to forcibly replace the pixel block with 0 in accordance with the control signal for replacement when the high frequency component of the pixel block in the image data is determined to be relatively small by the data processing unit. By replacing non-zero parts of some AC coefficients in the quantized discrete cosine transform coefficient Ruv with zeros, the data in those parts is truncated to reduce the amount of compressed data, and the data processing unit is used to reduce the amount of compressed data. When it is determined that the high frequency component of the Down conversion coefficient Ruv
Is output as it is.

In the image coding apparatus according to a fifth aspect, the data processing means detects an edge component of the image data for each pixel block, compares the edge component value with a threshold value, A data processing circuit that derives a replacement control signal based on the accumulated value of the comparison result is used.

In the image coding apparatus according to a sixth aspect, an edge detector for detecting an edge component of the image data for each of the pixel blocks, and a value of the edge component detected by the edge detector is set to a threshold value. A threshold magnitude comparator for comparing the magnitude with a threshold value from a setting register to derive a comparison result output when the value of the edge component is greater than the threshold value; and accumulating the frequency of the threshold comparison result output. A data processing circuit comprising a counter and comparing the cumulative value of the comparison result output frequency by the counter with a set value of a high frequency component number setting register to detect whether or not the high frequency component is large, as the data processing means. Things.

In the image encoding apparatus according to a seventh aspect, as the data processing means, an edge component of the image data is detected for each of the pixel blocks, and a value of the edge component is compared with a threshold value. In a device using a data processing circuit for deriving a data control signal based on a cumulative value of a comparison result, a plurality of the thresholds are set, and a plurality of different data are set based on a cumulative value of the comparison result related to each threshold. A control signal is derived, and an AC coefficient of the quantized discrete cosine transform coefficient Ruv of a different portion is forcibly set to 0 according to each of the plurality of data control signals.

[0038] In the image encoding apparatus according to an eighth aspect of the present invention, the data processing means detects an edge component of the image data for each pixel block, and sets a value of the edge component to a second threshold value and a second threshold value. A third different from the second threshold
And the second and third data control signals are calculated based on the cumulative value of the comparison result for the second threshold value and the cumulative value of the comparison result for the third threshold value. This uses a derived data processing circuit.

In an image encoding apparatus according to a ninth aspect, an edge detector for detecting an edge component of the image data for each of the pixel blocks, and a value of the edge component detected by the edge detector is used for a second and a second component. A threshold for comparing the second and third threshold values from the third threshold setting register with each other and deriving a comparison result output when the value of the edge component is larger than the second and third threshold values; Value comparator,
A second counter for accumulating the frequency of the comparison result output for the second threshold value, and a third counter for accumulating the frequency of the comparison result output for the third threshold value; And comparing the cumulative value of the output frequency of the comparison result by the second counter, the cumulative value of the output frequency of the comparison result by the third counter, and the set value of the high frequency component number setting register to detect whether or not the high frequency component is large. A data processing circuit is used as the data processing means.

In the image coding apparatus according to the tenth aspect,
As the data processing means, an edge component of the image data is detected for each of the pixel blocks, and a value of the edge component is set to a fourth threshold value, a fifth threshold value different from the fourth threshold value, and A cumulative value of the comparison result for the fourth threshold value and a cumulative value of the comparison result for the fifth threshold value, which are compared with a sixth threshold value different from the fourth and fifth threshold values. And a data processing circuit that derives fourth, fifth, and sixth data control signals based on the cumulative value of the comparison result of the sixth threshold value.

In the image coding apparatus according to the eleventh aspect,
An edge detector for detecting an edge component of the image data for each of the pixel blocks, and a value of the edge component detected by the edge detector being set to a fourth value from a fourth, fifth, and sixth threshold setting registers. , Fifth and sixth threshold values, and the edge component values are compared with the fourth, fifth and sixth threshold values.
A threshold magnitude comparator for deriving a comparison result output when the threshold value is larger than a threshold value, a fourth counter for accumulating the frequency of the comparison result output for the fourth threshold value, and a fifth counter. A fifth counter for accumulating the frequency of the comparison result output with respect to the threshold value, and a sixth counter for accumulating the frequency of the comparison result output with respect to the sixth threshold value; The cumulative value of the comparison result output frequency, the cumulative value of the comparison result output frequency by the fifth counter, the cumulative value of the comparison result output frequency by the sixth counter, and the set value of the high frequency component number setting register. A data processing circuit for comparing and detecting whether there are many high frequency components is used as the data processing means.

[0042]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing the overall configuration of an image encoding device according to the present invention. In the figure, 1 is an input terminal for inputting image data of an 8 × 8 pixel block, 2 is a discrete cosine transform unit that performs a discrete cosine transform (DCT), 3 is a zigzag transform unit, 4 is a quantizing unit,
Is a quantization table, 6 is an entropy coding unit, 7 is a coding table, 8 is an output terminal for outputting parameters and code data, 9 is an edge detector, and processes discrete cosine transform (DCT) coefficients after quantization. This is a data processing means including a data processing circuit.

FIG. 2 shows an embodiment in which an edge component of image data is detected from an input terminal for inputting image data of an 8 × 8 pixel block according to the first embodiment of the present invention. FIG. 9 is a diagram showing a circuit configuration of a data processing circuit 9 that generates a signal for performing effective code amount control based on the number of comparison results with the data processing circuit 9. In the figure, 10
Is an edge detector, 11 is a first magnitude comparator, 12 is a threshold setting register, 13 is a counter, 14 is a high frequency component number setting register, and 15 is a second magnitude comparator. In the embodiment according to the present invention, the basic configuration and operation other than those related to the data processing circuit 9 are shown in FIG. 5 and subsequent figures. Same as the one. However, the processing order of the zigzag conversion unit 3 and the quantization unit 4 is changed so that the processing is performed in the zigzag conversion unit 3 after the processing by the quantization unit 4.

Next, the operation of the image coding apparatus according to the present invention will be described with reference to FIGS. The image data input from the input terminal 1 is converted into an 8 ×
A two-dimensional discrete cosine transform is executed for each component image Pxy composed of eight pixels, and the resulting 64 discrete cosine transform coefficients Suv are sent to the quantization unit 4. In the quantization unit 4, an operation of dividing the discrete cosine transform coefficient Suv by the value Quv in the quantization table 5 is performed, and the result is sent to the zigzag conversion unit 3. The data processed by the zigzag converter 3 according to the control signal from the data processing circuit 9 is encoded by the entropy encoder 6. This is a schematic description of the operation in the first embodiment according to the present invention.

Next, the operation will be described in more detail.
First, as image data, from the image data input terminal 1,
64 having an 8-bit width, for example, constituting an image block
Component images Pxy (x, y = 0, 1, 2, 2,
3..., 7) are input. Here, the order xy of each component image Pxy is 2 in the image data.
It shows the position of the dimensional matrix. For example, x =
2, y = 1, the component image Pxy is a two-dimensional matrix row position (x = 0, 1, 2, 3,...) In the image data as an image block constituting the image data.
, 7) is 2 and the column position (y = 0, 1, 2, 3,...,
7) shows one.

The 64 component images Pxy of the input image data are sent to the discrete cosine transform unit 2. In the discrete cosine transform unit 2, the divided 8 × 8
A two-dimensional discrete cosine transform (DCT) is performed for each component image Pxy composed of pixel blocks.

As a result of the two-dimensional discrete cosine transform (DCT) in the two-dimensional discrete cosine transform unit 2, 64 data having the order uv corresponding to the order xy of the component image Pxy as the data of the component image Pxy for the 8 × 8 pixel block (= 8 × 8) discrete cosine transform coefficients Suv are obtained. Here, each discrete cosine transform coefficient S
The order uv of uv is the order x of each component image Pxy
It corresponds to y and indicates a two-dimensional matrix position in the image data. For example, the discrete cosine transform coefficient Suv in the case of u = 2, v = 1 is a two-dimensional matrix row position (u = 0, 1, 2, 3,..., 7) in the image data as an image block constituting the image data. ) Is 2
, Where the row position (v = 0, 1, 2, 3,..., 7) is 1.

The obtained 64 discrete cosine transform coefficients S
uv is sent from the discrete cosine transform unit 3 to the quantization unit 4. The 64 discrete cosine transform coefficients Suv have different step sizes for each coefficient position using the quantization table 5, that is, different step sizes for each two-dimensional position on the image data corresponding to the order uv of the coefficient Suv. , Are quantized by the quantization unit 4.

The 64 quantized discrete cosine transform coefficients Ruv output from the quantizing section 4 and sent to the zigzag transform section 3 and arranged in the serial order are converted from the serial order by the zigzag transform section 3 to the zigzag transform section 3. Can be reordered.

The 64 quantized discrete cosine transform coefficients Ruv output from the zigzag transformer 3 are sent to the entropy encoder 6. In the entropy encoder 6, the quantized discrete cosine transform coefficient Ruv is stored in the encoding table 7.
And the encoded data is output from the output terminal 8 in units of several bytes (for example, 16-bit width).

On the other hand, the image data input from the input terminal 1 is also sent to the data processing circuit 9. In the data processing circuit 9, the input image data is sent to the edge detector 10. The edge detector 10 performs Laplacian mask processing on the image data of 8 × 8 pixels, for example, every 3 × 3 pixels. When mask processing of 3 × 3 pixels is performed on 8 × 8 pixels, edge data of 6 × 6 pixels is obtained as a result. The obtained edge data of 6 × 6 pixels is sent to the first magnitude comparator 11 and compared with the value preset in the threshold setting register 12. If the edge data is larger than the first threshold, the counter 13
Is incremented by one.

The count value of the counter 13 as a result of the threshold value determination for 6 × 6 pixels is sent to the second magnitude comparator 15. The second magnitude comparator 15 compares the value of the counter 13 with the preset value of the high frequency component number setting register 14. When the value of the counter 13 is larger than the value of the high frequency component number setting register 14 set in advance, it is determined that the number of high frequency components of 8 × 8 pixels is large, and the first replacement data control signal is sent to the zigzag converter 3. Not issued. On the other hand, when the value of the counter 13 is not larger than the value of the high frequency component number setting register 14, it is determined that the high frequency component is small, and the first replacement data control signal is output to the zigzag converter 3.

In the zigzag converter 3, the data quantized by the quantizer 4 is rearranged from a serial order to a zigzag order. As for the 64 quantized discrete cosine transform coefficients Ruv rearranged in the zigzag order, the first data is a DC (direct current) coefficient, and the second to 64th is an AC (alternating current) coefficient. These coefficients are sent to the entropy encoding unit 6 as needed, and the entropy encoding unit 6 performs Huffman encoding processing.

Here, when the block data is rearranged in the zigzag order in the zigzag conversion section 3, a replacement data control signal from the second magnitude comparator 15 provided in the data processing circuit 9 constituting the data processing means is used. Out of 64 discrete cosine transform coefficients Ruv each having an AC coefficient arranged in a zigzag order, for example, the 36th one of a specific portion of the block data set in advance corresponding to the replacement data control signal. Thereafter, all the AC coefficient values of the 64th discrete cosine transform coefficient Ruv are replaced with “0”.

On the other hand, when the replacement data control signal is not output from the second magnitude comparator 15 provided in the data processing circuit 9 constituting the data processing means, that is, when it is determined that the high frequency component is large, A portion other than “0” in the AC coefficient of the discrete cosine transform coefficient Ruv is not replaced with “0”, and the AC coefficient is output as it is.

As described above, the output of the second magnitude comparator 15 allows the zigzag converter 3 to generate the discrete cosine transform coefficient R
Since the AC coefficient value of uv is forcibly set to “0” from a certain point, the amount of data entropy-encoded in the entropy encoding unit 6 at the next stage can be saved.
Then, the discrete cosine transform coefficient Ruv is quantized in consideration of the fact that the energy of the image is concentrated in the low-frequency region. As described above, the discrete cosine transform coefficient Suv is
It is quantized with a different step size for each coefficient position using the quantization table 5, that is, with a different step size for each two-dimensional position on the image data corresponding to the order uv of the coefficient Suv, and When the replacement data control signal is output from the second magnitude comparator 15 of the data processing circuit 9 because the high frequency component of the pixel block in the image data is small, the AC coefficient value of the discrete cosine transform coefficient Ruv is obtained. Can be forcibly set to "0" from a certain point, the visual degradation of the reproduced image can be reliably suppressed.

In the first embodiment, a discrete cosine transform unit for performing discrete cosine transform for each image block, a quantizing unit for performing linear quantization of the discrete cosine transformed coefficients at different step sizes for each coefficient position, A zigzag converter for zigzagging block data arranged in a serial order, a data processing circuit for generating a data control signal for each pixel block, and entropy encoding of the data-processed coefficients using a Huffman encoding scheme An image encoding device comprising an entropy encoding unit, which compresses image data, detects an edge component of the image data for each pixel block, compares the edge component value with a first threshold value, Based on the number of comparison results, the non-zero portion of the AC coefficient quantized by the quantization unit is output as it is or The data processing circuit for controlling whether a non-zero part of the converted AC coefficient is forcibly set to 0 and outputting the data, and reducing the amount of compressed data by truncating some non-zero parts of the AC coefficient It is intended to be.

In the first embodiment, the non-zero portion of the quantized AC coefficient is forcibly set to zero and output. At a predetermined location or later.

According to the first embodiment of the present invention, each component image Pxy (where the degree xy indicates a two-dimensional matrix position in the image data of each component image) constituting an image block of image data is provided. A discrete cosine transform is performed, and a discrete cosine transform coefficient Suv composed of coefficients having DC components or AC components arranged in a predetermined order (where the order uv indicates a two-dimensional matrix position in the image data of each component image Pxy) A discrete cosine transform unit 2 for deriving a two-dimensional matrix position on the image data corresponding to the order xy), and a discrete cosine transform comprising a coefficient having a DC component or an AC component derived by the discrete cosine transform unit 2 Coefficient Suv is converted to a two-dimensional matrix on the image data A quantizing unit 4 that performs linear quantization with a different step size for each unit to derive a quantized discrete cosine transform coefficient Ruv, and rearranges the quantized discrete cosine transform coefficient Ruv arranged in a predetermined order and in a serial order zigzag. A zigzag transformer 3 and an entropy encoder 6 for performing entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding scheme.
In an image encoding device that compresses image data, detects whether a high-frequency component of a pixel block in the image data is large and controls whether or not to output a replacement control signal, and controls the zigzag conversion unit. Quantized discrete cosine transform coefficients Ru arranged in zigzag
and v. a data processing circuit comprising a data processing circuit 9 for controlling whether or not AC coefficients after a predetermined specific coefficient among v are forcibly replaced with 0 according to the replacement control signal. 9 when the high-frequency component of the pixel block in the image data is determined to be relatively small by the data processing means comprising the non-zero part of some AC coefficients in the quantized discrete cosine transform coefficient Ruv by the replacement control signal. Is replaced by 0 to reduce the amount of compressed data by truncating the data in that portion, and when the data processing means including the data processing circuit 9 determines that the high frequency components of the pixel blocks in the image data are relatively large. Outputting the AC coefficient in the quantized discrete cosine transform coefficient Ruv as it is Since the so that, it is possible to make effective data compression amount, it is possible to obtain a picture coding apparatus that can reliably suppress the visual degradation of the reproduced image.

Embodiment 2 FIG. 3 is a block diagram showing an embodiment in which an edge component of image data is detected from an input terminal for inputting image data of an 8 × 8 pixel block according to a second embodiment of the present invention. FIG. 9 is a diagram showing a circuit configuration of a data processing circuit 9 that generates a signal for performing effective code amount control based on the number of comparison results with the threshold value of.

In FIG. 3, reference numeral 10 denotes an edge detector;
Is a first magnitude comparator, 14 is a high frequency component number setting register, 15 is a second magnitude comparator, 16 is a second threshold setting register, 17 is a third threshold setting register, 18
Is a second counter, and 19 is a third counter. In the second embodiment according to the present invention, data processing circuit 9
The configuration and operation other than the parts are the same as those of the first embodiment.
It is similar to that of

Next, the operation of the image coding apparatus according to the present invention will be described with reference to FIGS. First, the image data input from the input terminal 1 is converted into a discrete cosine transform unit 2
Is 2 for each 8 × 8 pixel component image Pxy
A dimensional discrete cosine transform is performed and the resulting 64
The discrete cosine transform coefficients Suv are sent to the quantization unit 4. In the quantization unit 4, an operation of dividing the discrete cosine transform coefficient Suv by the value Quv in the quantization table 5 is performed, and the result is sent to the zigzag conversion unit 3. The image data input from the input terminal 1 is also sent to the data processing circuit 9.

In the data processing circuit 9, the input image data is sent to the edge detector 10. Edge detector 10
For example, for image data of 8 × 8 pixels, for example, 3 × 3
The mask processing of the LaBlacian is executed for each pixel. 8x8
When mask processing of 3 × 3 pixels is performed on the pixels, edge data of 6 × 6 pixels is obtained as a result. The obtained edge data of 6 × 6 pixels is sent to the first magnitude comparator 11, and the second and third threshold values set in advance in the threshold value setting registers 16 and 17 are used. They are compared in size. If the edge data is larger than the second threshold and not larger than the third threshold, the value of the counter 18 is incremented by one. Further, when the edge data is larger than the third threshold value, the value of the counter 19 is counted up by one.

The count value of the counter 18 and the count value of the counter 19 as a result of the threshold value determination for 6 × 6 pixels are sent to the second magnitude comparator 15. The second magnitude comparator 15 compares the value of the counter 18 and the value of the counter 19 with the preset value of the high frequency component number setting register 14. When the value of the counter 19 is the maximum, this 8 ×
It is determined that the high frequency components of the eight pixels are large, and neither the second replacement data control signal nor the third replacement data control signal is output to the zigzag converter 3. Next, when the value of the counter 18 is the maximum, it is determined that the high frequency component of this 8 × 8 pixel is slightly large, and the second replacement data control signal is output to the zigzag converter 3. When the preset value of the high frequency component number setting register 14 is the maximum, it is determined that the high frequency component is small, and the third replacement data control signal is output to the zigzag converter 3.

In the zigzag transformer 3, the quantized discrete cosine transform coefficients Ruv quantized in the quantizer 4 are rearranged from serial order to zigzag order. Further, when a second replacement data control signal is output from the second magnitude comparator 15 and it is determined that the high frequency component of 8 × 8 pixels is slightly large, when rearranging in a zigzag order, In accordance with the second replacement data control signal from the second magnitude comparator 15, a specific portion of the quantized discrete cosine transform coefficient Ruv preset corresponding to the second replacement data control signal, that is, a zigzag 6 arranged in order
Of the four quantized discrete cosine transform coefficients Ruv, for example, all the AC coefficient values of the 48th to 64th quantized discrete cosine transform coefficients Ruv are forcibly set to “0”.

Further, the second magnitude comparator 15 outputs the second
Is output, and it is determined that the high-frequency component of 8 × 8 pixels is small, the third replacement data control signal from the second magnitude comparator 15 outputs the third replacement data control signal. Out of 64 quantized discrete cosine transform coefficients Ruv arranged in a zigzag order, for example, the 36th or later among the quantized discrete cosine transform coefficients Ruv preset in correspondence with Up to 64th quantized discrete cosine transform coefficient Ruv
Are forcedly set to “0”.

The 64 quantized discrete cosine transform coefficients Ruv rearranged in a zigzag order are a DC (DC) coefficient for the first data, and an AC (AC) coefficient for the second to 64th data. Becomes These quantized discrete cosine transform coefficients are sent to the entropy encoding unit 6 as needed, and the entropy encoding unit 6 performs Huffman encoding processing.

As described above, the value of the zigzag converter 3 is forcibly set to "0" from a certain point by the output of the second magnitude comparator 15, so that the entropy encoder 6 in the next stage
Can save the amount of data entropy-encoded. And the discrete cosine transform coefficient Ru
v is quantized in consideration of the fact that the energy of the image is concentrated in the low frequency region, and the second magnitude comparator 15 of the data processing circuit 9 controls the second replacement data control. When the signal is output, the high frequency component of the pixel block in the image data is slightly high, and when the third replacement data control signal is output, the high frequency component of the pixel block in the image data is low. Since it is time, A of discrete cosine transform coefficient Ruv
Even if the C coefficient value is forcibly set to “0” from a certain point, visual deterioration of the reproduced image can be surely suppressed.

In the second embodiment, a discrete cosine transform unit for performing discrete cosine transform for each image block, and a quantizer for linearly quantizing the discrete cosine transformed coefficients at different step sizes for each coefficient position, A zigzag converter for zigzagging block data arranged in a serial order, a data processing circuit for generating a data control signal for each pixel block, and entropy encoding of the data-processed coefficients using a Huffman encoding scheme In an image coding apparatus including an entropy coding unit for compressing image data, an edge component of the image data is detected for each pixel block, and the value of the edge component is set to a second threshold value and a third threshold value. The threshold value is compared with the threshold value. The data processing circuit for controlling whether the non-zero part of the quantized AC coefficient is output as it is or the non-zero part of the quantized AC coefficient is forcibly set to zero and output. This is to reduce the amount of compressed data by truncating the data in the non-zero portion of the coefficient.

According to the second embodiment of the present invention, an edge detector 10 for detecting an edge component of the image data for each pixel block, and a value of the edge component detected by the edge And compare the magnitude with the second and third thresholds from the third threshold setting registers 16 and 17 and output a comparison result output when the value of the edge component is larger than the second and third thresholds A threshold value comparator 11 to be derived, a second counter 18 for accumulating the frequency of the comparison result output for the second threshold value,
A third counter for accumulating the frequency of the comparison result output with respect to the third threshold value; and a cumulative value of the comparison result output frequency by the second counter and a third counter for accumulating the frequency. By using, as the data processing means, a data processing circuit that compares the cumulative value of the comparison result output frequency with the set value of the high frequency component number setting register 14 and detects whether the number of high frequency components is large,
When the value of the counter 19 is maximum, it is determined that the high frequency component of the pixel block in the pixel data is large, and when the value of the second counter is maximum, it is determined that the high frequency component of the pixel block in the pixel data is slightly high, When the value of the preset high frequency component number setting register 14 is the maximum, it is determined that the high frequency component of the pixel block in the pixel data is small, so it is more accurately determined whether the high frequency component of the pixel block in the pixel data is large. As a result, it is possible to obtain an image encoding device that can effectively compress the data amount and can surely suppress the visual deterioration of the reproduced image.

Embodiment 3 FIG. 4 shows an edge component of image data detected from an input terminal for inputting image data of an 8 × 8 pixel block according to the third embodiment of the present invention, and the value of the edge component, the fourth threshold value, and the fifth threshold value are detected. By the number of comparison results of the threshold value of the sixth threshold and the sixth threshold,
FIG. 3 is a diagram illustrating a circuit configuration of a data processing circuit 9 that generates a signal for performing effective code amount control. 4, reference numeral 10 denotes an edge detector, 11 denotes a first magnitude comparator, 14 denotes a high frequency component number setting register, 15 denotes a second magnitude comparator, 20 denotes a fourth threshold value setting register, and 21 denotes a fourth threshold value setting register. 5 is a threshold setting register, 22 is a sixth threshold register, 23 is a fourth counter, 24 is a fifth counter, and 25 is a sixth counter. In the third embodiment according to the present invention, the configuration and operation other than the data processing circuit 9 are the same as those in the first embodiment.

Next, the operation of the image coding apparatus according to the present invention will be described with reference to FIGS. First, the image data input from the input terminal 1 is converted into a discrete cosine transform unit 2
Is 2 for each 8 × 8 pixel component image Pxy
A dimensional discrete cosine transform is performed and the resulting 64
The discrete cosine transform coefficients Suv are sent to the quantization unit 4. In the quantization unit 4, the discrete cosine transform DCT coefficient Su
An operation of dividing v by the value Quv of the quantization table 5 is performed,
It is sent to the zigzag converter 3. The image data input from the input terminal 1 is also sent to the data processing circuit 9.

In the data processing circuit 9, the input image data is sent to the edge detector 10. Edge detector 10
For example, for image data of 8 × 8 pixels, for example, 3 × 3
The mask processing of the LaBlacian is executed for each pixel. 8x8
When mask processing of 3 × 3 pixels is performed on the pixels, edge data of 6 × 6 pixels is obtained as a result. The obtained edge data of 6.times.6 pixels is sent to the first magnitude comparator 11, and the fourth threshold value and the fifth threshold value set in advance in the threshold value setting registers 21, 22, 23 are set. Value and sixth
Is compared with the threshold value. If the edge data is larger than the fourth threshold and not larger than the fifth threshold, the value of the counter 23 is counted up by one. Further, when the edge data is larger than the fifth threshold value,
If it is not larger than the threshold value, the value of the counter 24 is counted up by one. Further, when the edge data is larger than the sixth threshold value, the value of the counter 25 is counted up by one.

The count value of the counter 23, the count value of the counter 24, and the count value of the counter 25 as a result of the threshold value determination for the 6 × 6 pixels are calculated by the second magnitude comparator 15.
Sent to In the second magnitude comparator 15, the counter 23
, The value of the counter 24, the value of the counter 25, and the value of the preset high frequency component number setting register 14 are compared in magnitude. When the value of the counter 25 is the maximum,
It is determined that the high-frequency component of × 8 pixels is very large, and neither the fourth replacement data control signal nor the fifth replacement data control signal or the sixth replacement data control signal is output to the zigzag converter 3. Next, when the value of the counter 24 is the maximum, this 8 ×
It is determined that the high frequency components of the eight pixels are not particularly large, and the fourth
To the zigzag conversion unit 3. Next, when the value of the counter 23 is the maximum, it is determined that the high frequency component of the 8 × 8 pixels is not so large, and the fifth replacement data control signal is output to the zigzag converter 3. And
When the preset value of the high frequency component number setting register 14 is the maximum, it is determined that the high frequency component is small, and the sixth replacement data control signal is output to the zigzag converter 3.

In the zigzag transformation unit 3, the quantized discrete cosine transform coefficients Ruv quantized in the quantization unit 4 are rearranged from a serial order to a zigzag order. Further, when a fourth replacement data control signal is output from the second magnitude comparator 15, and it is determined that the high frequency components of 8 × 8 pixels are not particularly large, the rearrangement in the zigzag order is performed. The quantized discrete cosine transform coefficient Ru set in advance corresponding to the fourth replacement data control signal by the fourth replacement data control signal from the second magnitude comparator 15
v, that is, among 64 quantized discrete cosine transform coefficients Ruv arranged in a zigzag order,
For example, all the AC coefficient values of the 58th to 64th quantized discrete cosine transform coefficients Ruv are forcibly set to “0”.

Further, the second magnitude comparator 15 to the fifth magnitude comparator 15
When it is determined that the high-frequency component of 8 × 8 pixels is not so large, the fifth replacement data control signal from the second magnitude comparator 15 outputs A specific part of the quantized discrete cosine transform coefficients Ruv preset corresponding to the fifth replacement data control signal, that is, arranged in a zigzag order 6
Of the four quantized discrete cosine transform coefficients Ruv, for example, all the AC coefficient values of the 48th to 64th quantized discrete cosine transform coefficients Ruv are forcibly set to “0”.

The second magnitude comparator 15 to the sixth magnitude comparator 15
Is output, and it is determined that the high-frequency component of 8 × 8 pixels is small, the sixth replacement data control signal from the second magnitude comparator 15 determines the sixth replacement data control signal. Out of 64 quantized discrete cosine transform coefficients Ruv arranged in a zigzag order, for example, the 36th or later among the quantized discrete cosine transform coefficients Ruv preset corresponding to the replacement data control signal Up to 64th quantized discrete cosine transform coefficient Ruv
Are forcedly set to “0”.

The 64 quantized discrete cosine transform coefficients Ruv rearranged in the zigzag order are as follows. Becomes These quantized discrete cosine transform coefficients are sent to the entropy encoding unit 6 as needed, and the entropy encoding unit 6 performs Huffman encoding processing.

As described above, the value of the zigzag converter 3 is forcibly set to "0" from a certain point by the output of the second magnitude comparator 15, so that the entropy encoder 6 in the next stage
Can save the amount of data entropy-encoded. And the discrete cosine transform coefficient Ru
v is quantized in consideration of the fact that the energy of the image is concentrated in the low-frequency region, and the second replacement data control by the fourth magnitude comparator 15 of the data processing circuit 9 is performed. When the signal is output, the high frequency component of the pixel block in the image data is not particularly large, and when the fifth replacement data control signal is output, the high frequency component of the pixel block in the image data is output. Is not too large, and the sixth replacement data control signal is output, because the high frequency component of the pixel block in the image data is small, and the AC coefficient value of the discrete cosine transform coefficient Ruv is However, even if it is forcibly set to "0", the visual degradation of the reproduced image can be surely suppressed.

In the third embodiment, a discrete cosine transform unit for performing a discrete cosine transform for each image block, a quantizing unit for performing a linear quantization on the discrete cosine transformed coefficients with a different step size for each coefficient position, A zigzag converter for zigzagging block data arranged in a serial order, a data processing circuit for generating a data control signal for each pixel block, and entropy encoding of the data-processed coefficients using a Huffman encoding scheme In an image coding apparatus having an entropy coding unit for compressing image data, an edge component of the image data is detected for each pixel block, and a value of the edge component is set to a fourth threshold value and a fifth threshold value. The threshold value is compared with the sixth threshold value, and the number of comparison results of the fourth threshold value and the number of comparison results of the fifth threshold value are calculated. Serial sixth Accept outputs the partial non-zero AC coefficients quantized by the quantization unit based on the number of comparison results of the threshold of the quantized AC
The data processing circuit controls whether the non-zero part of the coefficient is forcibly set to zero and outputs the data, and reduces the amount of compressed data by truncating some non-zero part of the AC coefficient. .

According to the third embodiment of the present invention, an edge detector 10 for detecting an edge component of the image data for each pixel block, and a value of the edge component detected by the edge , Fifth and sixth threshold setting registers 20, 21, and 22 from the fourth and fifth threshold setting registers.
A threshold magnitude comparator 11 for comparing the magnitude with the threshold value and the sixth threshold value and deriving a comparison result output when the value of the edge component is greater than the fourth, fifth and sixth threshold values; A fourth counter 23 for accumulating the frequency of the comparison result output for the fourth threshold value, and a fifth counter 24 for accumulating the frequency of the comparison result output for the fifth threshold value
And a sixth counter 25 for accumulating the frequency of the comparison result output with respect to the sixth threshold value, wherein the cumulative value of the comparison result output frequency by the fourth counter 23 and the fifth counter 24, the cumulative value of the comparison result output frequency by the sixth counter 25, and the set value of the high frequency component number setting register 14 to detect whether or not the high frequency component is large. By using the data processing circuit 9 as the data processing means, when the value of the sixth counter 25 is maximum, it is determined that the high frequency component of the pixel block in the pixel data is extremely large, and the value of the fifth counter 24 is determined. Is maximum, it is determined that the high frequency components of the pixel block in the pixel data are not particularly large, and when the value of the fourth counter 23 is maximum, High-frequency component of the unit block is determined to be not too much, because if the value of the high frequency component number setting register 14 is the maximum set in advance was set to determine the high frequency component of the pixel block in the pixel data is small,
An image encoding device that can more accurately and precisely determine whether or not a pixel block has a large number of high-frequency components in pixel data, can effectively compress the amount of data, and can reliably suppress visual degradation of a reproduced image. Can be obtained.

[0082]

According to the first aspect of the present invention, a discrete cosine is used for each component image Pxy (where the order xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data. The transform is performed, and a discrete cosine transform coefficient Suv composed of coefficients having DC components or AC components arranged in a predetermined order (where the order uv is an order xy indicating a two-dimensional position in the image data of each component image Pxy) A discrete cosine transform unit for deriving a corresponding two-dimensional position on the image data), and a discrete cosine transform coefficient Suv composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit. Performs linear quantization with a different step size for each of the above two-dimensional positions, and performs quantization discrete cosine transformation. A quantization unit for deriving the coefficients Ruv,
A zigzag transform unit that arranges the quantized discrete cosine transform coefficients Ruv arranged in a predetermined order and arranged in a serial order zigzag; An image coding apparatus for compressing image data, comprising detecting whether or not a high-frequency component of a pixel block in the image data is large and controlling whether or not to output a replacement control signal; Data processing means for controlling whether the AC coefficient of a preset portion of the discrete cosine transform coefficient Ruv is forcibly replaced with 0 according to the control signal for replacement, and a pixel in the image data by the data processing means; When it is determined that the high frequency component of the block is relatively small, the replacement control is performed. By replacing non-zero parts of some AC coefficients in the quantized discrete cosine transform coefficients Ruv with 0, the data in those parts is truncated to reduce the amount of compressed data, and the data processing means reduces the amount of compressed data. When it is determined that the high frequency component of the pixel block is relatively large, the AC coefficient in the quantized discrete cosine transform coefficient Ruv is output as it is, so that it is possible to effectively compress the data amount, and furthermore, to reproduce the reproduced image. It is possible to obtain an image encoding device that can reliably suppress visual deterioration.

According to the second aspect of the present invention, discrete cosine transform is performed for each component image Pxy (where the order xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data. And a discrete cosine transform coefficient Suv composed of coefficients having a DC component or an AC component arranged in a predetermined order (where the order uv corresponds to the order xy indicating a two-dimensional position in the image data of each component image Pxy). A discrete cosine transform unit for deriving the two-dimensional position on the image data) and a discrete cosine transform coefficient Suv composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit. Quantized discrete cosine transform by performing linear quantization with a different step size for each two-dimensional position A quantizer for deriving a number Ruv,
A zigzag transform unit for arranging the quantized discrete cosine transform coefficients Ruv arranged in a predetermined order and arranged in a serial order zigzag; and an entropy code for performing entropy encoding on the quantized discrete cosine transform coefficients Ruv using a Huffman coding method. An image coding apparatus for compressing image data, comprising detecting whether or not a high-frequency component of a pixel block in the image data is large, and controlling whether or not to output a replacement control signal. Data processing means for controlling whether AC coefficients subsequent to a predetermined specific coefficient among the quantized discrete cosine transform coefficients Ruv are forcibly replaced with 0 according to the control signal for replacement, It was determined that the high frequency component of the pixel block in the image data was relatively small At the same time, the non-zero part of some AC coefficients in the quantized discrete cosine transform coefficient Ruv is replaced with zero by the replacement control signal, thereby cutting off the data of the part to reduce the amount of compressed data and performing the data processing. When it is determined by the means that the high frequency component of the pixel block in the image data is relatively large, the AC coefficient in the quantized discrete cosine transform coefficient Ruv is output as it is, so that effective data amount compression can be performed. It is possible to obtain an image encoding device that can surely suppress visual deterioration of a reproduced image.

According to the third aspect of the present invention, discrete cosine transform is performed for each component image Pxy (where the degree xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data. And a discrete cosine transform coefficient Suv composed of coefficients having a DC component or an AC component arranged in a predetermined order (where the order uv corresponds to an order xy indicating a two-dimensional position in the image data of each component image Pxy). A discrete cosine transform unit for deriving a two-dimensional position on the image data) and a discrete cosine transform coefficient Suv composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit. Quantized discrete cosine transform by performing linear quantization with a different step size for each two-dimensional position A quantizer for deriving a number Ruv,
A zigzag transform unit that arranges the quantized discrete cosine transform coefficients Ruv arranged in a predetermined order and arranged in a serial order zigzag; An image coding apparatus for compressing image data, comprising detecting whether a high-frequency component of a pixel block in the image data is large and controlling whether or not to output a replacement control signal, and performing the zigzag conversion. A part of the quantized discrete cosine transform coefficients Ruv arranged in zigzag by the unit is forced to 0 in accordance with the replacement control signal.
A data processing means for controlling whether or not to replace the quantized discrete cosine transform coefficient by the replacement control signal when the data processing means determines that the high frequency component of the pixel block in the image data is relatively small. By replacing non-zero parts of some AC coefficients in Ruv with 0, the data in those parts is truncated to reduce the amount of compressed data, and the data processing means has relatively large high-frequency components of pixel blocks in the image data. When it is determined that the AC coefficient in the quantized discrete cosine transform coefficient Ruv is output as it is, an effective data amount compression can be performed, and furthermore, an image which can surely suppress the visual deterioration of the reproduced image can be obtained. An encoding device can be obtained.

According to the fourth aspect, the discrete cosine transform is performed for each component image Pxy (where the degree xy indicates the two-dimensional position of each component image in the image data) constituting the image block of the image data. And a discrete cosine transform coefficient Suv composed of coefficients having a DC component or an AC component arranged in a predetermined order (where the order uv corresponds to the order xy indicating a two-dimensional position in the image data of each component image Pxy). A discrete cosine transform unit for deriving the two-dimensional position on the image data) and a discrete cosine transform coefficient Suv composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit. Quantized discrete cosine transform by performing linear quantization with a different step size for each two-dimensional position A quantizer for deriving a number Ruv,
A zigzag transform unit that arranges the quantized discrete cosine transform coefficients Ruv arranged in a predetermined order and arranged in a serial order zigzag; An image coding apparatus for compressing image data, comprising detecting whether a high-frequency component of a pixel block in the image data is large and controlling whether or not to output a replacement control signal; Data for controlling whether the AC coefficients after a predetermined coefficient among the quantized discrete cosine transform coefficients Ruv rearranged in zigzag by the unit are forcibly replaced with 0 according to the control signal for replacement. Processing means, and a pixel block in the image data is provided by the data processing means. When it is determined that the high frequency component is relatively small, the non-zero part of some AC coefficients in the quantized discrete cosine transform coefficient Ruv is replaced with 0 by the replacement control signal, thereby cutting off the data of that part. Since the amount of compressed data is reduced, and when the data processing unit determines that the high frequency component of the pixel block in the image data is relatively large, the AC coefficient in the quantized discrete cosine transform coefficient Ruv is output as it is. In addition, it is possible to obtain an image encoding device that can effectively compress the data amount and can surely suppress the visual deterioration of the reproduced image.

In the image encoding apparatus according to a fifth aspect, the data processing means detects an edge component of the image data for each pixel block, compares the edge component value with a threshold value, Since a data processing circuit that derives a replacement control signal based on the accumulated value of the comparison result was used,
To obtain an image coding apparatus capable of accurately determining whether a pixel block has a large number of high-frequency components in pixel data, effectively compressing the amount of data, and reliably suppressing visual deterioration of a reproduced image. Can be.

According to the sixth aspect, an edge detector for detecting an edge component of the image data for each of the pixel blocks, and a value of the edge component detected by the edge detector is supplied from a threshold setting register. A threshold value comparator for comparing the threshold value with the threshold value to derive a comparison result output when the value of the edge component is larger than the threshold value; and a counter for accumulating the frequency of the threshold comparison result output. Since the data processing circuit for comparing the cumulative value of the comparison result output frequency by the counter with the set value of the high frequency component number setting register and detecting whether the high frequency component is large is used as the data processing means, the pixel data It is possible to accurately determine whether there is a large amount of high-frequency components in the pixel block in, effectively compress the amount of data, and ensure visual deterioration of the reproduced image. It is possible to obtain a picture coding apparatus can be suppressed.

According to the seventh aspect, the data processing means detects an edge component of the image data for each pixel block, compares the value of the edge component with a threshold value, and accumulates the comparison result. In a device using a data processing circuit that derives a data control signal based on a value, a plurality of the thresholds are set, and a plurality of different data control signals are derived based on an accumulated value of a comparison result related to each threshold. Then, A of the quantized discrete cosine transform coefficient Ruv of a different part according to each of the plurality of data control signals
Since the C coefficient is forcibly set to 0, it is possible to accurately determine whether or not there is a large number of high frequency components in the pixel block in the pixel data, and to effectively compress the data amount. It is possible to obtain an image encoding device capable of reliably suppressing the target deterioration.

According to the eighth aspect, as the data processing means, an edge component of the image data is detected for each pixel block, and the value of the edge component is set to a second threshold value and the second threshold value. Comparing the threshold value with a third threshold value different from the threshold value, based on the cumulative value of the comparison result for the second threshold value and the cumulative value of the comparison result for the third threshold value. Since the data processing circuit for deriving the data control signal of No. 3 is used, it is possible to more accurately determine whether or not the pixel block contains a large number of high frequency components in the pixel data, and to effectively compress the data amount. It is possible to obtain an image encoding device that can surely suppress visual deterioration of an image.

According to the ninth aspect, an edge detector for detecting an edge component of the image data for each of the pixel blocks, and a value of the edge component detected by the edge detector is set to a second and a third value. A threshold magnitude comparator for comparing the magnitude of the edge component with the second and third thresholds from the threshold value setting register and deriving a comparison result output when the value of the edge component is greater than the second and third thresholds And a second counter for accumulating the frequency of the comparison result output for the second threshold value, and a third counter for accumulating the frequency of the comparison result output for the third threshold value, A cumulative value of the comparison result output frequency by the second counter, a cumulative value of the comparison result output frequency by the third counter,
Since a data processing circuit that compares the value set in the high frequency component number setting register and detects whether the number of high frequency components is large is used as the data processing means, it is more accurate to determine whether the pixel block in the pixel data has many high frequency components. In addition, it is possible to obtain an image encoding device which can discriminate carefully and can effectively compress the data amount and can surely suppress the visual deterioration of the reproduced image.

According to the tenth aspect, as the data processing means, an edge component of the image data is detected for each pixel block, and the value of the edge component is set to a fourth threshold value and the fourth threshold value. A fifth threshold value different from the threshold value and a sixth threshold value different from the fourth and fifth threshold values, and a cumulative value of a comparison result of the fourth threshold value and the fifth threshold value are compared. Since the data processing circuit that derives the fourth, fifth, and sixth data control signals based on the cumulative value of the comparison result for the threshold value and the cumulative value of the comparison result for the sixth threshold value is used. An image code that can more accurately and precisely determine whether or not a pixel block has a large number of high-frequency components in pixel data, can effectively compress the amount of data, and can reliably suppress visual deterioration of a reproduced image. It can be obtained device.

According to the eleventh aspect, an edge detector for detecting an edge component of the image data for each of the pixel blocks, and a value of the edge component detected by the edge detector is set to a fourth, fifth, and fifth values. The threshold value is compared with the fourth, fifth, and sixth threshold values from the sixth threshold setting register, and if the value of the edge component is larger than the fourth, fifth, and sixth threshold values, the comparison result is output. , A fourth counter for accumulating the frequency of the comparison result output for the fourth threshold value, and accumulating the frequency of the comparison result output for the fifth threshold value And a sixth counter that accumulates the frequency of the comparison result output for the sixth threshold value, wherein the cumulative value of the comparison result output frequency by the fourth counter is 5 by the counter A data processing circuit that compares the accumulated value of the comparison result output frequency, the accumulated value of the comparison result output frequency by the sixth counter, and the set value of the high frequency component number setting register to detect whether or not the high frequency component is large; Since it is used as the data processing means, it is possible to more accurately and precisely determine whether or not the pixel data has a large number of high-frequency components in the pixel data, and to effectively compress the data amount. , An image encoding apparatus capable of surely suppressing the above can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an overall configuration of an image encoding device according to the present invention.

FIG. 2 is a diagram illustrating a circuit configuration of a data processing circuit that performs effective code amount control according to the first embodiment of the present invention.

FIG. 3 is a diagram showing a circuit configuration of a data processing circuit for performing effective code amount control according to a second embodiment of the present invention;

FIG. 4 is a diagram showing a circuit configuration of a data processing circuit for performing effective code amount control according to a third embodiment of the present invention.

FIG. 5 is a basic block illustrating a conventional image encoding device.

FIG. 6 is a diagram illustrating a distribution graph of quantized DCT coefficients after two-dimensional discrete cosine transform and quantization of an 8 × 8 pixel block in the image encoding device.

FIG. 7 is a block diagram illustrating a DC coefficient grouping circuit in the image encoding device.

FIG. 8 is a diagram illustrating the concept of DC coefficient difference processing in the image encoding device.

FIG. 9 is a diagram showing a grouping table for performing grouping of DC coefficients in the image encoding device.

FIG. 10 is a block diagram illustrating an AC coefficient grouping circuit in the image encoding device.

FIG. 11 is a diagram illustrating a zigzag scan order of an 8 × 8 pixel block in the image encoding device.

FIG. 12 is a diagram illustrating a grouping table for performing grouping of AC coefficients in the image encoding device.

FIG. 13 shows a DC coefficient and A in an image encoding apparatus.
It is a block diagram which shows Huffman coding of C coefficient.

FIG. 14 is a diagram illustrating a code table for performing differential coding of DC coefficients in the image coding device.

FIG. 15 shows AC coefficients and D in an image coding apparatus.
It is a figure showing the procedure in which C coefficient is grouped.

FIG. 16 is a diagram illustrating a code table for encoding AC coefficients in the image encoding device.

[Explanation of symbols]

1 input terminal, 2 discrete cosine transform unit (DCT), 3
A zigzag conversion unit, a 4 quantization unit, a 5 quantization table,
6 entropy coding section, 7 coding table, 8
Output terminal, 9 data processing circuit, 10 edge detector,
11 large / small comparator, 12 threshold setting register, 13
Counter, 14 high frequency component number setting register, 15
Large / small comparator, 16, 17 threshold setting register, 1
8, 19 counter, 20, 21, 22 threshold setting register, 23, 24, 25 counter, 60 grouping unit, 61 block delay unit, 62 DC differentiator, 63
Grouping unit, 65 one-dimensional Huffman coding unit, 66
DC code table section, 67 DC additional bit combining section, 6
8 Coupling circuit for coupling DC coded signal and AC coded signal, 70 Huffman coding section, 86 Ruv signal input terminal, 92 determination section, 93 run length counter, 9
4 grouping unit, 95 two-dimensional Huffman coding unit, 9
6 AC encoding unit, 97 AC additional bit combining unit.

Continued on the front page F term (reference) 5C059 MA00 MA23 MC01 MC11 MC14 MC22 MC32 MC34 MC38 ME02 ME08 PP16 PP22 TA47 TA49 TB08 TC04 TC10 TC43 TD07 TD12 UA02 5C078 AA04 BA21 BA57 CA21 DA00 DA01 DB07 DB18 9A001 EE02 EE04 GG31H27

Claims (11)

[Claims] 1. A discrete cosine transform is performed for each component image Pxy (here, an order xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data, and the components are arranged in a predetermined order. Discrete cosine transform coefficients Suv composed of arranged coefficients having DC components or AC components (where the order uv is 2 on the image data corresponding to the order xy indicating a two-dimensional position in the image data of each component image Pxy). And a discrete cosine transform coefficient S composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit.
a quantizer for linearly quantizing uv at different step sizes for each two-dimensional position on the image data to derive a quantized discrete cosine transform coefficient Ruv, and the quantized discrete cosine arranged in a predetermined order and arranged in a serial order Conversion coefficient Ru
a zigzag transformation unit that rearranges v in a zigzag manner, and an entropy encoding unit that performs entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding method, and an image encoding device that compresses image data. Detecting whether or not a high-frequency component of a pixel block in the image data is large and controlling whether or not to output a replacement control signal, the AC coefficient of a predetermined portion of the quantized discrete cosine transform coefficient Ruv Data processing means for controlling whether to forcibly replace the pixel data with 0 in response to a replacement control signal, wherein the replacement is performed when the data processing means determines that the high frequency component of the pixel block in the image data is relatively small. Some AC coefficients in the quantized discrete cosine transform coefficient Ruv by the control signal for By replacing the non-zero part with 0, the data in that part is truncated to reduce the amount of compressed data, and when the data processing unit determines that the high frequency component of the pixel block in the image data is relatively large, the quantum An image coding apparatus, wherein an AC coefficient in a generalized discrete cosine transform coefficient Ruv is output as it is. 2. A discrete cosine transform is performed for each component image Pxy (where the order xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data, and the components are arranged in a predetermined order. Discrete cosine transform coefficients Suv composed of arranged coefficients having DC or AC components (where the order uv is 2 on the image data corresponding to the order xy indicating the two-dimensional position in the image data of each component image Pxy). And a discrete cosine transform coefficient S composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit.
a quantizer for linearly quantizing uv at different step sizes for each two-dimensional position on the image data to derive a quantized discrete cosine transform coefficient Ruv, and the quantized discrete cosine arranged in a predetermined order and arranged in a serial order Conversion coefficient Ru
a zigzag transformation unit that rearranges v in a zigzag manner, and an entropy encoding unit that performs entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding method. Detecting whether a high-frequency component of a pixel block in the image data is large and controlling whether or not to output a control signal for replacement is performed, and a predetermined specific coefficient among the arranged quantized discrete cosine transform coefficients Ruv is set. Data processing means for controlling whether the subsequent AC coefficient is forcibly replaced with 0 in accordance with the replacement control signal is provided. When the quantized discrete cosine transform coefficient Ruv is changed by the replacement control signal when When the non-zero part of some AC coefficients is replaced with zero, the data in that part is truncated to reduce the amount of compressed data. An image coding apparatus characterized in that when discriminated, the AC coefficient in the quantized discrete cosine transform coefficient Ruv is output as it is. 3. A discrete cosine transform is performed for each component image Pxy (where the order xy indicates a two-dimensional position of each component image in the image data) constituting an image block of image data, and the components are arranged in a predetermined order. Discrete cosine transform coefficients Suv composed of arranged coefficients having DC or AC components (where the order uv is 2 on the image data corresponding to the order xy indicating the two-dimensional position in the image data of each component image Pxy). And a discrete cosine transform coefficient S composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit.
a quantizer for linearly quantizing uv at different step sizes for each two-dimensional position on the image data to derive a quantized discrete cosine transform coefficient Ruv, and the quantized discrete cosine arranged in a predetermined order and arranged in a serial order Conversion coefficient Ru
a zigzag transformation unit that rearranges v in a zigzag manner, and an entropy encoding unit that performs entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding method, and an image encoding device that compresses image data. The quantized discrete cosine transform, which is arranged to detect whether or not a high-frequency component of a pixel block in the image data is large and to control whether or not to output a replacement control signal, is arranged in a zigzag manner by the zigzag converter. Data processing means is provided for controlling whether the AC coefficient of a preset portion of the coefficient Ruv is forcibly replaced with 0 in accordance with the replacement control signal, and the data processing means controls a high frequency of a pixel block in the image data. When it is determined that the component is relatively small, the replacement control signal By replacing non-zero parts of some AC coefficients in the generalized discrete cosine transform coefficient Ruv with zeros, the data in those parts is truncated to reduce the amount of compressed data, and the data processing means performs high-frequency conversion of pixel blocks in the image data. An image coding apparatus characterized in that when it is determined that the number of components is relatively large, the AC coefficient in the quantized discrete cosine transform coefficient Ruv is output as it is. 4. A discrete cosine transform is performed for each component image Pxy (where the order xy indicates a two-dimensional position of each component image in the image data) constituting an image block of the image data, and the components are arranged in a predetermined order. Discrete cosine transform coefficients Suv composed of arranged coefficients having DC components or AC components (where the order uv is 2 on the image data corresponding to the order xy indicating a two-dimensional position in the image data of each component image Pxy). And a discrete cosine transform coefficient S composed of a coefficient having a DC component or an AC component derived by the discrete cosine transform unit.
a quantizer for linearly quantizing uv at different step sizes for each two-dimensional position on the image data to derive a quantized discrete cosine transform coefficient Ruv, and the quantized discrete cosine arranged in a predetermined order and arranged in a serial order Conversion coefficient Ru
a zigzag transformation unit that rearranges v in a zigzag manner, and an entropy encoding unit that performs entropy encoding on the quantized discrete cosine transform coefficient Ruv using a Huffman encoding method. Quantized discrete cosine transform coefficients arranged to be arranged in a zigzag manner by the zigzag conversion unit by detecting whether or not a high frequency component of a pixel block in the image data is large, and controlling whether or not to output a replacement control signal. A data processing unit for controlling whether AC coefficients after a predetermined coefficient among Ruv are forcibly replaced with 0 according to the control signal for replacement, and a pixel block in the image data by the data processing unit. When it is determined that the high-frequency component is relatively small, the replacement control signal Serial quantized discrete cosine transform coefficients Ru
By replacing non-zero parts of some AC coefficients in v with 0, the data in those parts is truncated to reduce the amount of compressed data, and the data processing means has relatively large high-frequency components of pixel blocks in the image data. Wherein the AC coefficient in the quantized discrete cosine transform coefficient Ruv is output as it is. 5. The data processing unit detects an edge component of the image data for each pixel block, compares the value of the edge component with a threshold value, and performs replacement based on a cumulative value of the comparison result. 5. The image encoding device according to claim 1, wherein a data processing circuit for deriving a control signal is used. 6. An edge detector for detecting an edge component of the image data for each of the pixel blocks, and comparing a value of the edge component detected by the edge detector with a threshold value from a threshold value setting register. A threshold magnitude comparator that derives a comparison result output when the value of the edge component is greater than the threshold value; and a counter that accumulates the frequency of the threshold comparison result output. 3. A data processing circuit for comparing the cumulative value of the result output frequency with a set value of a high frequency component number setting register to detect whether or not the high frequency component is large is used as the data processing means. Item 5. The image encoding device according to any one of Items 4. 7. The data processing means detects an edge component of the image data for each pixel block, compares the value of the edge component with a threshold value, and performs data control based on a cumulative value of the comparison result. In the one using a data processing circuit for deriving a signal, a plurality of the thresholds are set,
Deriving a plurality of different data control signals based on the accumulated value of the comparison results relating to the respective thresholds, and calculating the AC coefficients of the quantized discrete cosine transform coefficients Ruv of different portions according to each of the plurality of data control signals 5. The image encoding apparatus according to claim 1, wherein is set to 0 forcibly. 8. The data processing means detects an edge component of the image data for each pixel block, and sets a value of the edge component to a second threshold value and a third threshold value different from the second threshold value. And the second and third data control signals are calculated based on the cumulative value of the comparison result for the second threshold value and the cumulative value of the comparison result for the third threshold value. The image encoding device according to claim 7, wherein a data processing circuit for deriving the image is used. 9. An edge detector for detecting an edge component of the image data for each of the pixel blocks, and a value of the edge component detected by the edge detector from a second and third threshold value setting registers. A threshold magnitude comparator for comparing the magnitude with the second and third thresholds to derive a comparison result output when the value of the edge component is greater than the second and third thresholds; A second counter for accumulating the frequency of the comparison result output for the threshold value, and a third counter for accumulating the frequency of the comparison result output for the third threshold value
And comparing the cumulative value of the comparison result output frequency by the second counter, the cumulative value of the comparison result output frequency by the third counter, and the set value of the high frequency component number setting register. 8. The image coding apparatus according to claim 7, wherein a data processing circuit for detecting whether a high frequency component is large is used as said data processing means. 10. The data processing means detects an edge component of the image data for each pixel block, and sets a value of the edge component to a fourth threshold value and a fifth threshold value different from the fourth threshold value. And a sixth threshold value different from the fourth and fifth threshold values, and the cumulative value of the comparison result for the fourth threshold value and the sixth threshold value for the fifth threshold value are compared. A data processing circuit for deriving fourth, fifth, and sixth data control signals based on a cumulative value of a comparison result and a cumulative value of a comparison result for the sixth threshold value is used. Item 8. The image encoding device according to Item 7. 11. An edge detector for detecting an edge component of the image data for each pixel block, and setting a fourth, fifth and sixth threshold values of the edge component detected by the edge detector. A threshold value which compares the magnitude with the fourth, fifth and sixth threshold values from the register and derives a comparison result output when the value of the edge component is greater than the fourth, fifth and sixth threshold values A magnitude comparator, a fourth counter for accumulating the frequency of the comparison result output for the fourth threshold value, and a fifth counter for accumulating the frequency of the comparison result output for the fifth threshold value. A sixth counter for accumulating the frequency of the comparison result output with respect to the sixth threshold value, wherein the accumulated value of the comparison result output frequency by the fourth counter is compared with the comparison value by the fifth counter. Cumulative result output frequency A data processing circuit for comparing the product value, the accumulated value of the comparison result output frequency by the sixth counter, and the set value of the high frequency component number setting register to detect whether or not the high frequency component is large, as the data processing means; The image encoding device according to claim 7, wherein the image encoding device is used.